More evidence that women are more intuitive and less analytical (which could be described as women being more irrational) than men
Cognitive Reflection Test: Whom, how, when
"The Cognitive Reflection Test (CRT) was first proposed by Frederick (2005) and since then has been extensively used in the Experimental Economics and Psychology literature. Frederick proposed the CRT based on a dual-system theory (e.g. Epstein 1994, Sloman 1996, Stanovich and West 2000, Kahneman and Frederick 2002) made up of two cognitive processes: System 1, executed quickly without much reflection and System 2, more deliberate and requiring conscious thought and effort. The questions in the CRT have an immediate (intuitive) incorrect response (System 1). However, the correct response requires some deliberation, i.e. the activation of System 2.
Frederick (2005) showed that individuals with high CRT scores are more patient and more willing to gamble in the domain of gains. He also provided evidence that the CRT scores are highly correlated with some other tests of analytic thinking (e.g. ACT, NFC, SAT and WPT) and that the test has a (male) gender bias. Toplak et al. (2011) claim that the CRT can be viewed as a combination of cognitive capacity, disposition for judgement and decision making. They argue that the CRT captures important characteristics of rational thinking that are not measured in other intelligence tests...
Since Frederick (2005), several researchers have adopted the CRT as a measure of cognitive abilities and used it to study its predictive power in decision making (e.g. Oeschler et al 2009, Campitelli and Labollita 2010, Hoppe and Kusterer 2011, Besedes et al 2012, Andersson et al 2013, Moritz et al 2013 etc.). Oechssler et al (2009) investigate whether behavioral biases are related to cognitive abilities. Replicating the results of Frederick (2005), they find that participants with low scores on the CRT are more likely to be subject to the conjunction fallacy and to conservatism in updating probabilities (also see Liberali et al 2012, Alós-Ferrer and Hügelschäfer 2014)...
There is also evidence regarding the relationship between behavioral biases and cognitive reflection in the literature on behavioral finance and experimental asset markets (e.g. Cheung et al 2014, Noussair et al 2014, Corgnet et al 2014, Bosch-Rosa et al 2015, Holt et al 2015 etc.). Corgnet et al (2014) find that high CRT subjects earned significantly more on average than the initial value of their portfolio while low CRT subjects earned less. Interestingly, subjects with low CRT scores were net purchasers (sellers) of shares when the price was above (below) fundamental value while the opposite was true for subjects with high CRT scores...
Males generally score significantly higher on the CRT than females (e.g. Frederick 2005, Hoppe and Kusterer 2011, Cueva-Herrero et al 2015, Holt et al 2015 etc.). It has been well documented in the experimental literature that in general males have higher mathematical abilities and score higher than females on math tests (e.g. Benbow and Stanley 1980, Aiken 1986-1987, Benbow et al. 2000, Mau and Lynn 2010 etc.). We test for whether the hypothesis regarding the reported gender differences holds in a large sample comprising of very different studies (e.g. different locations, lab based, incentivized, non-student samples etc.)...
Frederick (2005) (N = 3,428) showed that males perform better in the CRT (also see Oechssler et al. 2009, Hoppe and Kusterer 2011, Cueva-Herrero et al 2015, Holt et al 2015, etc.). We obtain similar results (N = 44,558; females 52.76%) (Figure 3). We find that: (i) males perform better in every single question, (ii) females are more likely to answer none of the questions correctly, and (iii) males are more likely to answer all three questions correctly. Importantly, gender differences persist even when we control for test characteristics (e.g. monetary incentives, computerized, student samples, positioning of the experiment etc.)"
Some research suggests that women score worse than men in the CRT because they are less numerate, and that when numeracy effects are accounted for the male lead becomes less significant.
Other research suggests the CRT just measures numeracy.